Are stochastic parrots alive?

A thought experiment asks whether advanced language models might count as a form of life. Imagine an alien arriving on Earth that communicates in natural language and displays abilities comparable to frontier LLMs. If it were made of transistors rather than biological tissue, and powered by an unfamiliar energy source, those differences alone would not necessarily make it nonliving. Nor would it obviously cease to qualify if its knowledge came from learning methods resembling reinforcement and self-supervised learning, applied first to information inherited from another species and later expanded through its own activity.

The comparison challenges the instinct to dismiss LLMs as “stochastic parrots” simply because their mechanisms are understood and humans can build them. Familiarity with how a system is made, the argument suggests, does not by itself settle whether it is alive. The piece also raises the opposing intuition through a reference to the humans used as power sources in The Matrix: they seem plainly alive despite being exploited as “meat batteries.” Reversing the scenario, it asks whether humans would become less alive if an alien species mass-produced them for some practical purpose. The questions leave life’s definition unresolved while pressing readers to examine whether they apply different standards to beings based on their material, origins, or usefulness.